Open Data in Practice

Zachary Batist

McGill University
Epidemiology, Biostatistics and Occupational Health

October 28, 2025

How do researchers actually engage with open science practices, principles and tooling?

vs.

How advocates claim science works, and data’s role in their vision of scientific knowledge production

Open Science: a movement to make scientific research more accessible to scientists and society in general.

Open Data: the application of open science principles to data

About me

About me
What are data?
How open science tends to imagine data
Examples
Take-Aways
Questions?

Primary Interests

Collaborative research practices
How digital tools and infrastructures reconfigure knowledge production
Specifically:
- Infrastructures that support data sharing
- Open source communities of practice
- The cultural and epistemic implications of research infrastructures and policy

How does open science (esp. open data) reconfigure — or attempt to reconfigure — collaborative experiences?
What values are associated with open data, and how do they intersect with scientists actual needs, desires and values?

Currently:

Technical and administrative systems that scaffold the Covid Immunity Task Force Databank

Previously (and ongoing, to a lesser extent):

Data management within archaeological projects
Archaeological data sharing, integration and reuse

Also:

Collaborative software development practices
Small-scale and community-driven data sharing initiatives

What are data?

About me
What are data?
How open science tends to imagine data
Examples
Take-Aways
Questions?

Descriptive accounts of observed phenomenon
Evidence that forms the basis of a claim
Means of communicating observations from one set of circumstances to another

Cats in my neighbourhood:
Name	Colour	Sex	Feral
Ellie	Mix	F	Y
Bob	Brown	M	N
Charlie	Brown	M	N
Jasper	Black	M	N
????	Tortie	?	Y
Spencer	White	M	N

Descriptive accounts of observed phenomenon
Evidence that forms the basis of a claim
Means of communicating observations from one set of circumstances to another

Data were collected to document Ellie’s dietary preferences, by comparing the quantity of food consumed of various flavours.

Based on analysis of this data, we found that Ellie has a clear preference for chicken pate relative to all other options presented to her.

Descriptive accounts of observed phenomenon
Evidence that forms the basis of a claim
Means of communicating observations from one set of circumstances to another

Dearest Colleagues,

I’m enclosing the data from my neighbourhood observations.

My observations were done between 2PM-6PM, during my Sunday strolls.

It includes cats I saw in ground-floor apartment windows, and cats I encountered on the street.

Note that I saw no outdoor cats on rainy days, and I skipped my walk last week.

Feel free to make use of this very important information as you see fit.

What are data?

Descriptive accounts of observed phenomenon
Evidence that forms the basis of a claim
Means of communicating observations from one set of circumstances to another

What are data?

Descriptive accounts of observed phenomenon
Evidence that forms the basis of a claim
Means of communicating observations from one set of circumstances to another

Implies:

Data are fixed and stable
Data are generated prior to and separate from analysis
Aura of authority, stability, and truthfulness.

What are data?

Descriptive accounts of observed phenomenon
Evidence that forms the basis of a claim
Means of communicating observations from one set of circumstances to another

Implies:

Data are created under specific circumstances
Data are created for targeted purposes
Data are informed by partial background knowledge

At the same time, conceiving of data as means of communicating observations between specific circumstances adds a layer of subjectivity.
It acknowledges that data are not universal, that they’re created for targeted purposes, and that they’re limited by partial knowledge and material factors.

This is especially important in a curatorial context, since it presents data as a product of human creativity.
And like all things made my people, they carry traces of the social and material contexts of their creation.

As digital curators, a significant part of our job is ensuring that the information contained in these records can be understood by those who seek to read them.
And so our role is that of a mediator, connecting the perspectives of data creators and designated communities who will be accessing the data using the knowledge that each party already has.

How open science tends to imagine data

About me
What are data?
How open science tends to imagine data
Examples
Take-Aways
Questions?

The Open Data Imaginary

Non-material and non-political
Reflections of natural reality
Infinitely re-configurable

The Open Data Imaginary

Non-material and non-political
Reflections of natural reality
Infinitely re-configurable

Implies:

Spreadsheets, databases and articles are considered value-neutral representations
Belief that anyone can create and access data
- BUT: Requires scientific resources, internet access, computational resources, requisite expertise to make sense of data

So first off, there is a strong tendency in the open science imaginary to conceive of data as non-material.
That is, the accumulation and assembly of data is thought to contribute to a species-level understanding of the world, which is not held by any one individual but is stored as a form of common knowledge.
While there is some acknowledgement that information is stored in physical media like books, scientific reports and internet-connected archives, these documents are often perceived of as value-neutral and purely representational in nature.
However, not everyone can create data, and standards for validating the legitimacy of data are rooted in material constraints.
For instance, if I work for a small or under-funded university, I may be unable to afford equipment that enables me to generate data at the same scale or calibre as my peers at Harvard or Oxford.
Moreover, you need internet access, computational resources and knowledge of how to make sense of data in order for data to be useful, and if you’re not thinking about non-English speakers or people without internet access, you’re not actually making things universally accessible.

The Open Data Imaginary

Non-material and non-political
Reflections of natural reality
Infinitely re-configurable

Implies:

Belief that data are reflections of reality, rather than outcomes of decisions and actions
Data’s legitimacy considered to derive from their objectivity

The Open Data Imaginary

Non-material and non-political
Reflections of natural reality
Infinitely re-configurable

Implies:

Diverse data can click together to form new knowledge
More data ⟹ more possible configurations
Ignores the extreme challenges involved in making heterogeneous datasets compatible

Science is intrinsically material and positional.

Open science fails when it does not account for these things.

Examples

About me
What are data?
How open science tends to imagine data
Examples
Take-Aways
Questions?

Database of Obsidian Sourcing Studies (DObsiSS)

Different geological sources identified by different labs (as of 1998, ~27 years ago)

Data presented as prose, and mixed with analysis of non-chemical characteristics (2003)

dataARC

Graphic overview of data sources embedded in dataARC (https://data-arc.org)

Take-Aways

About me
What are data?
How open science tends to imagine data
Examples
Take-Aways
Questions?

Take-Aways

Open data as currently practiced encourages taking data at face-value
Most researchers intuitively understand data to be more complex than that
More valuable insights may be obtained by
- Smaller-scale initiatives
- Directed by specific objectives
- With specific designated communities in mind

Questions?

About me
What are data?
How open science tends to imagine data
Examples
Take-Aways
Questions?